{
"$schema": "https://json-schema.org/draft/2020-12/schema",
"$id": "https://toroid.fyi/schema/instance.schema.json",
"title": "Obfuscratic Instance Register - instance record",
"description": "One dated event drawn from a pre-registered source-exhaustive frame. An instance is NOT an architecture. It carries no five-signature score and no composite. It carries a direction sign, a cluster assignment, an evidence tag, and its provenance.",
"type": "object",
"required": [
"instance_id", "frame_id", "source_ref", "event_date", "locus_id",
"stated_beneficiary_class", "realized_incidence_class", "direction",
"evidence_tag", "coded_by", "coded_on", "provenance"
],
"additionalProperties": false,
"properties": {
"instance_id": {
"type": "string",
"pattern": "^OIR-[0-9]{4}-[0-9]{6}$",
"description": "OIR--. Assigned at ingest, never reused, never reassigned."
},
"frame_id": {
"type": "string",
"description": "The pre-registered frame this instance was drawn from. An instance with no frame_id may not enter any statistical test."
},
"frame_position": {
"type": "integer",
"minimum": 1,
"description": "Ordinal position of this item in the frame's enumeration, before any coding. Present so a reader can verify the frame was walked exhaustively rather than sampled by salience."
},
"source_ref": {
"type": "object",
"required": ["citation", "url", "retrieved_on"],
"additionalProperties": false,
"properties": {
"citation": {"type": "string", "description": "Document-level citation sufficient to locate the primary record without the URL."},
"url": {"type": "string", "format": "uri"},
"doc_identifier": {"type": "string", "description": "Federal Register document number, GAO report number, OIG report number, docket number, or equivalent."},
"retrieved_on": {"type": "string", "format": "date"},
"archive_url": {"type": "string", "format": "uri", "description": "Web archive snapshot. Required for any source that is not a permanent government document identifier."}
}
},
"event_date": {
"type": "string",
"format": "date",
"description": "Date on the primary record. Not the date of reporting about it."
},
"epoch": {
"type": "string",
"pattern": "^[0-9]{4}-Q[1-4]$",
"description": "Derived from event_date. Used by the temporal layer."
},
"locus_id": {
"type": "string",
"description": "Independent decision locus. THE UNIT OF STATISTICAL INDEPENDENCE. Two instances sharing a locus_id contribute jointly, not separately. See codebook section 3 for the locus rule."
},
"locus_rationale": {
"type": "string",
"description": "Why this locus and not a broader or narrower one. Required because locus assignment sets effective sample size and is therefore the most attackable coding decision in the register."
},
"stated_beneficiary_class": {
"$ref": "#/$defs/beneficiary_class",
"description": "Read off the authorizing document's own statement of purpose. Not inferred."
},
"stated_beneficiary_quote": {
"type": "string",
"description": "Verbatim purpose language from the primary document. Paraphrase is not permitted here."
},
"realized_incidence_class": {
"$ref": "#/$defs/beneficiary_class",
"description": "Where first-order incidence actually fell, per the evidence anchored in incidence_basis."
},
"incidence_basis": {
"type": "string",
"enum": ["agency_own_figures", "gao_or_oig_finding", "court_record", "third_party_measurement", "contemporaneous_reporting", "not_yet_established"],
"description": "What establishes realized incidence. 'not_yet_established' forces direction to 0 and excludes the instance from the test while retaining it in the register."
},
"direction": {
"type": "integer",
"enum": [-2, -1, 0, 1, 2],
"description": "Signed divergence of realized incidence from stated beneficiary. NEGATIVE VALUES MUST BE REACHABLE OR THE TEST IS VOID. -2 strongly counter-directional, -1 counter-directional, 0 symmetric or indeterminate, +1 directional, +2 strongly directional. See codebook section 4."
},
"direction_rationale": {"type": "string"},
"payoff_conjunction": {
"type": "boolean",
"description": "Whether the realized incidence class carries an identifiable financial-capture or political-alignment payoff. Per Part V, directional coding without payoff conjunction is noise rather than signature; recorded separately so the test can be run both ways."
},
"evidence_tag": {
"type": "string",
"enum": ["occurred", "projected", "contested", "correlation"],
"description": "Part III discipline. Only 'occurred' instances may anchor a realized-outcome claim. 'projected' instances are registered and excluded from the directional test."
},
"contestation": {
"type": "array",
"items": {
"type": "object",
"required": ["class", "summary", "source"],
"additionalProperties": false,
"properties": {
"class": {"type": "string", "enum": ["high_quality", "low_quality_motivated"], "description": "Per Part III: methodological critique, disinterested alternative measurement, timeframe or base-rate correction, scope or attribution fix = high_quality and adjusts confidence. Bare denial by an interested party, reframing without counter-evidence, ad hominem, asserted intent = low_quality_motivated, logged and carrying no weight."},
"summary": {"type": "string"},
"source": {"type": "string"},
"effect_on_direction": {"type": "integer", "enum": [-2, -1, 0, 1, 2], "description": "Revised direction if this contestation is accepted. Populated so the whole register can be re-run under the hostile reading."}
}
}
},
"architecture_ref": {
"type": ["string", "null"],
"description": "Optional link to a scored architecture in the register. An instance need not belong to any architecture, and belonging to one confers no score on the instance."
},
"coded_by": {"type": "string", "description": "Human coder identifier. A machine may not appear here."},
"coded_on": {"type": "string", "format": "date"},
"second_coder": {"type": "string", "description": "Second independent human coder, where the item was double-coded."},
"second_coder_direction": {"type": "integer", "enum": [-2, -1, 0, 1, 2]},
"provenance": {
"type": "object",
"required": ["proposed_by", "ratified"],
"additionalProperties": false,
"properties": {
"proposed_by": {
"type": "string",
"enum": ["human", "machine_extraction"],
"description": "How the candidate record entered the queue."
},
"proposing_agent": {"type": "string", "description": "Model or tool identifier where proposed_by is machine_extraction. Carried for the life of the record."},
"ratified": {
"type": "boolean",
"description": "False until a human has read the primary source and confirmed every field. UNRATIFIED RECORDS ARE EXCLUDED FROM EVERY STATISTICAL TEST AND FROM EVERY RENDERED FIGURE."
},
"ratified_by": {"type": "string"},
"ratified_on": {"type": "string", "format": "date"},
"machine_fields_altered_at_ratification": {
"type": "array",
"items": {"type": "string"},
"description": "Which machine-proposed field values the human changed. The running rate of alteration is the register's measured machine-extraction error rate and is published."
}
}
},
"notes": {"type": "string"}
},
"$defs": {
"beneficiary_class": {
"type": "string",
"description": "Closed vocabulary. Coders may not invent classes; a class that does not fit is a codebook amendment request, dated and logged.",
"enum": [
"general_public",
"program_enrolled_population",
"low_income_households",
"medically_dependent_population",
"veterans",
"federal_workforce",
"state_or_local_government",
"small_business",
"large_incumbent_firms",
"specific_contractor_or_vendor",
"financial_asset_holders",
"high_income_households",
"detention_or_enforcement_contractors",
"extractive_resource_licensees",
"named_individual_or_family",
"political_alignment_network",
"foreign_state_or_entity",
"no_identifiable_beneficiary",
"indeterminate"
]
}
}
}